| blob | 6989291d3105fce6ffaeeacb0dda95001f51458f |
1 This version of PsN is tested with perl-5.8.8
2 Depending on your distribution you may have to install the following
3 perl packages:
5 Math::Random
6 Storable::Store
8 All available from CPAN ( www.cpan.org )
10 Changelog
11 ----------------------------------------------
13 Changes from 2.2.4 to 2.2.5
15 Features
17 - mcs script renamed to
18 - <new script> can now run the basic model, unless allready run.
19 - <new script> calculates lots of nice results.
21 Bugs
23 - Modelfit did not copy extra subroutine files in some cases.
25 - <new script> Add IGNORE=@ to estimation models unless it
26 exists. Necessary for ignoring headers in table output.
28 - Memory leak in output parser plugged. Output files from simulations
29 could use vast amounts of memory.
31 - Bootstrap options "skip_minimization_terminated" can now be disabled
32 by typing "-no-skip-minimization_terminated" (or -noskip-...)
34 - cdd xv can no be disabled with -no-xv or -noxv
36 - mc tool no uses considerably less memory. It is a lot slower
37 because it reads a lot of unecessary stuff from disk, it should be
38 reviewed.
40 - Memory leak in modelfit due to intermediate "raw_results.csv" in
41 NM_runX directory fixed by removing the feature, should be removed.
43 - A initial estimate of form: $OMEGA BLOCK SAME would not be
44 recognized by PsN model file parser. This is now handled
45 correctly. In 2.2.X it can be avoided by typing BLOCK(X) SAME,
46 where X is the size of the block.
48 Changes from 2.2.3 to 2.2.4
50 Features
52 - "--compute_cwres" option renamed to simply "--cwres"
54 - Simulated data suitable for creating mirror plots can be generated
55 by using a new options called --mirror_plots. The option takes a
56 value which defines the number of simulations that will be
57 geneated.
59 - A new option, --iofv, enables the computation of individual
60 objective function values. These values are printed to a table file
61 called iotab*, where * is the same number as is defined for any
62 sdtab or patab tables. These names correspond to the normal Xpose
63 table file format. If no sdtab or patab table is found in the
64 NONMEM control stream, the iotab file is given number 1.
66 - msfo resumes. Basic support for automatically rerunning a model
67 that has the MSFO=msffile option on $ESTIMATION. If the option is
68 set, the "msffile" exists and the option --msfo_resume is given,
69 PsN will remove initial estimates, and add the $MSFI record. This
70 is very untested and if you have OMEGA blocks, the model will run
71 but PsN will be unable to read the output file properly.
73 - "no_remote_compile" and "no_remote_execution" options has been
74 removed, as they were quite seldomly used and added a lot of
75 complexity to parallization code.
77 - Added --prepend_model_file_name option which adds the model file
78 name to output files to prevent them from being overwritten.
80 - A new option called "crash_restarts" which controls the number of
81 times PsN will restart a crashed run. (Simply put, run is
82 classified as crashed if the output file looks like it is cut off)
84 - PsN will now automatically run R-scripts generated for the
85 bootstrap, cdd and llp if R is installed and configured.
87 - When the CONT data item is used - either defined by the user or
88 automatically through the -wrap_data option - the tables get a
89 weird format where the real table content is printed on CONT=0 rows
90 and nonsense data printed on CONT=1 rows. A new option called
91 -unwrap_table_files reformats the table files and removes the CONT
92 column as well as all CONT=1 rows.
94 - update_inits script that will put final estimates from a NONMEM
95 output file and put into a given model file. This is quite
96 experimental. PsN will reformat the model file quite lot and
97 comments in the file might get lost. Your model file will be copied
98 with the addition of a ".org" extension, so you wont loose your
99 model.
101 Bug Fixes
103 - SGE queue and resource options were ignored. Fixed thanks to Jeroen
104 Elassaiss and Stefan Verhoeven.
106 - LSF Monitor fixed to work with the new serialized structure of PsN,
107 again thanks to Jeroen and Stefan.
109 - CWRES table file number now defaults to 1 if there is no patab or
110 sdtab to take the number from.
112 - Fixed run bugs under Windows 2000.
114 - Better support for G77 under windows.
116 - Parser for output files improved
118 Changes from 2.2.2 to 2.2.3
120 Features
122 - There has been a lot of requests for a raw_results file for the
123 runs in the NM_run catalogs, i.e. a summary of the runs with
124 perturbed initial estimates. This is now implemented. One
125 raw_results file is now created in each NM_run* catalog.
127 - The R-scripts for graphical visualization of bootstrap and cdds
128 have been updated. The format for the graphs is now pdf instead of
129 postscript. The bootstrap script can now be adjusted to exclude
130 bootstrap samples that contain copies of a given set of
131 individuals.
133 - When the wrap_data option is used, the secondary columns, i.e. the
134 data columns that are kept on rows with CONT=1, are now available
135 in $ERROR.
137 Bug Fixes
139 - The calculation of CWRES was previously done on the _last_ run of a
140 set of retries, not on the _best_ run. This has now been corrected.
142 - The restarting of crashed runs was broken. This is now fixed.
144 - The asynchronuous job scheduling broke the handling of seed numbers
145 for random sequences (e.g. perturbation of initial estimates). This
146 is now fixed. The present version is however not compatible with
147 older versions and old runs should therefore not be resumed using
148 this version.
150 - wrap_data should now work as it should. The functionality was
151 broken as a consequence of the move to the asynchronuous job
152 handling.
154 - Labels for omegas were not handled correctly for SAME blocks. Among
155 other things, this had the effect that the updating of initial
156 estimates from the original model to the method created models
157 (e.g. bootstrap samples) was corrupted. This is now fixed.
159 - The transformation of the estimate of the omega matrix from
160 variance and covariances to standard deviations and correlations
161 was rewritten to address problems with a combination of BLOCK(X)
162 ans SAME where X was larger than 1.
164 - The selection of a 'best' model fit after a series of perturbations
165 of the initial estimates will be changed. In 2.2.2, the lowest OFV
166 among fits of the perturbed models was collected across all models
167 (still only comparing identical models with perturbed initial
168 estimates). In 2.2.3 the lowest OFV is collected in three
169 categories of model fits: those that minimized succesfully, those
170 that did not minimize successfully but returned an OFV plus an
171 estimate of the number of significant digits and finally failed
172 runs that managed to return an OFV but not much more. This will
173 give PsN a more fine-grained control over the results. For example,
174 it would be desirable to be able to return the run with the lowest
175 OFV within the selection of runs with a successful minimization
176 even though the lowest OFV of all perturbed runs was returned from
177 a run that terminated.
179 - The sigmas were not parsed correctly when tables were printed to
180 the output file. This is not fixed.
182 - Terminated covariance steps were sometimes still being parsed (and
183 failed). Fixed.
185 - Setting the clean level to 3 did not remove the
186 NM_run-directories. This is now fixed.
188 - The parsing of options for the initial values of omegas and sigmas
189 sometimes wrongly recognized letters behind a semi-colon (i.e. a
190 comment) as keywords for a block (FIX, SAME, etc). This is now
191 corrected.
193 - If you "fix" more than one omega or sigma in a block PsN would put
194 a FIX after each value, which NONMEM doesn't allow. Now PsN will
195 only put one FIX inside a block. However, if a any part of the
196 block has been fixed, its likely so that PsN can not(!) unfix
197 it.(so, one bug fixed, another introduced)
199 - If you an initial estimate was added with "_init_attr", the option
200 "add_if_absent" was true and the last initial estimate was "SAME"
201 PsN would try to add the estimate after "SAME" which is wrong, now
202 PsN adds a new record.
204 - Fixed a parsing error of the lst file. If simulation without a
205 SIGMA was performed, the parser failed to read OMEGA initial
206 estiamtes.
208 - The internal method "indexes" would not properly handle "SAME" in
209 initial estimates. The result was that "update_inits" would skip
210 initial values, and get them out of order. Fixed.
213 Changes from 2.2.2-beta4 to 2.2.2
215 Features
217 - CWRES can now be calculated automatically if you have R and Xpose
218 installed. Remember to add a definition of IPRED in your NONMEM model.
220 - This version handles non-PsN standard (or absent) suffixes for
221 model files better than earlier versions.
223 Bug Fixes
225 - MSFO files are now copied back to the same directory of the model
226 file.
228 - MSFI files are now automatically copied to the PsN NONMEM run
229 directories.
231 - Output files from runs where the $MSFI record was used were not
232 parsed correctly. This is now fixed.
234 - Output files from runs where no $OMEGA _or_ $SIGMA were defined
235 were not parsed correctly.
237 - The header and the results from problematic runs, i.e. failed
238 covariance step, warnings etc, were not handled correctly when the
239 raw_results files were compiled. This is now fixed.
241 - In 2.2.1, we added a check for a '1' to find the end of the omega
242 and sigma estimate areas. this was a mistake since problems with
243 more than 13 omegas or sigmas get a '1'-line inserted before the
244 14:th estimate (for some reason). This is now removed.
246 Changes from 2.2.2-beta3 to 2.2.2-beta4
248 Features
250 - No new features
252 Bug Fixes
254 - Adaptive number of threads is now working in the new asynchronuous
255 job handling
257 - The CWRES computation had some fixes
259 - $INFN is now printed before $MODEL
261 - $OMEGA is now printed befor $PK
263 - PRIOR is now handled in the output module. It is still a bit
264 problematic to parse PRIOR output files correctly. Use with
265 causion.
267 - The initial estimates of large omega and sigma matrices is now
268 better handled
270 - The order of parsing events in the output module is now working
271 better.
273 - The handling of retries (or the results of retries) is now working
274 as it should.
276 Changes from 2.2.2-beta2 to 2.2.2-beta3
278 Features
280 - CWRES: Conditional weighted residuals. The necessary verbatim code
281 for the computation of CWRES are now automatically added if the
282 option -compute_cwres is used.
284 - PRIOR: The use of priors in NONMEM VI is now supported in the
285 computer intensive methods
287 - The behavior of the parallel execution has been revised, moving
288 from a synchronuous to an asynchronuous solution. Previously, the
289 memory foot-print of a parallel execution of say 100 NONMEM jobs
290 was huge (counted in several GB) and did not scale well. Now, it
291 will very seldom exceed 100 MB and more or less keep that size,
292 regardless of the number of parallel jobs.
294 Bug Fixes
296 - PsN couldn't parse IGNORE=(ITEM=VALUE) options in $DATA. This is
297 (again) fixed. It was previously corrected in the development
298 branch of PsN but never transferred to the stable branch.
300 - Fixed SAME block correlation computation.
302 Changes from 2.2.2-beta1 to 2.2.2-beta2
304 Features
306 - The parsing of the NONMEM output files has been revised and a
307 strict check is now performed after each section has been parsed. A
308 parsing error is raised if any non-compliant structures are
309 found. By doing this we get a better idea of which runs that really
310 finishes and which runs that are pre-maturely terminated
311 (e.g. through a operating system error). Remember that this is a
312 beta release and that there could still be valid text structures
313 that are not yet recognised correctly by PsN.
315 - Support for United Devices grip MP has been added and is available using
316 "-run_on_ud" command line option.
318 Bug Fixes
320 - the options directory and backwards_dir can now be specified in the
321 scm configuration file
323 - It is now possible to skip the definition of either continuous and
324 categorical covariates in the scm configuration file
326 - The exponential relation is now added correctly in the
327 scm. Previously it was incorrectly added as (for weight on volume)
328 TVV = THETA(1)*(1+EXP(THETA(2)*(WGT-median(WGT))))
329 Now it is added as
330 TVV = THETA(1)*EXP(THETA(2)*(WGT-median(WGT)))
332 - The output parse now computes correlations for omegas and sigmas correctly
333 when you have SAME blocks.
335 Changes from 2.2.1 to 2.2.2-beta1
337 Features
339 - The number of significant digits per parameter after a terminated
340 minimization are now parsed.
342 Bug Fixes
344 - The sample size option of the bootstrap did not work correctly with
345 stratified resampling. This is now fixed.
348 Changes from 2.2.0 to 2.2.1
350 Features
352 - Added support for the Intel Fortran compiler (mostly for those
353 using the new Intel Macs).
355 - We have added two new model fit quality terms: 'terminated' and
356 'really_bad' to the previous 'normal' and 'picky'.
358 picky - Minimization and covariance steps successful, no
359 parameters near boundaries and no warnings from the
360 covariance step.
362 normal - Minimization successful
364 terminated - Estimates of the objective function value and the
365 number of significant digits are available.
367 really_bad - An estimate of the OFV is available (but usually not
368 much more...).
370 Given that more than one run from a run has been performed (with
371 different initial estimates but otherwise identical) the following
372 selection procedure is done:
374 The lowest OFV of all tries is collected. Then, all runs within
375 5 units of the lowest OFV is tried according to
377 1. Of the runs passing the picky criteria, choose the run with the
378 highest number of significant digits.
380 2. If no run has passed the criteria above: of the runs passing the
381 normal criteria, choose the one with the highest number of
382 sginificant digits.
384 3. If no run has passed the criteria above: of the runs passing the
385 terminated criteria, choose the one with the highest number of
386 sginificant digits.
388 4. If still no run has passed the criteria above: of all runs that
389 at least have an estimate of the OFV, choose the run with the
390 lowest OFV.
392 5. As a last resort, choose the first run.
394 Comments: This is of course a very pragmatic approach but it is
395 needed to make the automatic procedures of the PsN tools to
396 work. We cannot force the user to manually scrutinise the results
397 of every single run of for example a bootstrap. The user has the
398 opportunity to do so if he or she wishes. If you do not feel
399 comfortable with this selection procedure or if you have comments
400 on how to improve it, please contact me (Lars Lindbom) at
401 lars.lindbom (at) farmbio.uu.se.
403 Bugfixes
405 - Data files with spaces around commas used for columns separation
406 could be problematic to parse. This is now fixed.
408 - There was a serious error in the seed handling in version
409 2.2.0. Under the following circumstances, the seed number was
410 ignored and a default seed number used: If the output file from the
411 original run was not present in the same directory as the original
412 model file when a PsN tool was started. This affected all PsN
413 tools, except the execute tool. This is now fixed.
415 - Up to 20 omegas can now be parsed in the output file. This is still
416 a limitation and it will be addressed in later releases.
418 - On fast distributed systems with a shared file system, the massive
419 parallelisation of the PsN tools could cause synchronization
420 problems (files could sometimes not be found, read or written). The
421 root of this problem is somewhat outside of the scope of PsN but we
422 have tried to find a workaround for this for the LSF grid system.
424 Changes from 2.1.10 to 2.2.0 ( released for PsN course )
426 Features
428 - PsN Output module use a proper Matrix library for handling some of
429 the matrices of the outputfile.
431 - The bootstrap can now be resumed with a lower number of
432 samples. Meaning that if you are in a hurry and want to stop the
433 bootstrap and get results after 1000 out of 2000 runs, you can. At
434 a later time you can then resume the 2000 runs.
436 - Nonmem warnings are now printed if "--verbose" is givven.
438 - Automatic incrementation of directory names are a bit smarter (no
439 more "directory modelfit_dirX is in the way" messages).
441 - Clean is much more stringent and intelligent and now manages to
442 remove entire tool directories if desired.
444 - option list printed from "-h|-?" only shows utility specific options.
446 - Default values configurable on a per utility basis.
448 - Online documentation improved
450 - Parallel run of NONMEM is not aborted as before if a outputfile is
451 not correctly parsed. It is just ignored, as if it had not
452 minimized successfully. (actually it is the output parser that
453 doesn't die, but the effect is the same)
455 - Parallel run is a bit nicer to the system when compiling many
456 NONMEMS. By adding a configurable delay between starts.
458 - Matlab and R scripts for various plots from cdd and bootstrap
459 results are distributed with PsN.
461 - Vast improvments in results presentations from all tools.
463 - Improvements in rerun algoritm with new "min_retries" option.
465 - Output from running utilties are much nicer.
467 - new "-quick_summary" prints minimization message and ofv value for
468 each finished run. As does "-verbose" option which also prints
469 "R:X" for each retrie (X is the retrie number).
471 - new "-summary" options prints diagnostics about the tool run when
472 it is finished. For "execute" the "-summary" option will have the
473 same function as the "sumo" utiltiy as well as some diagnostics.
475 - PsN now recognizes that control files may require fortran files and
476 copies them to the run directory. No need to add fotran files to
477 the --extra_files option any more.
479 - Added --extra_output which allows you to copy more than the output
480 file from the run directory.
482 Bugs
484 - PsN couldn't parse IGNORE=(ITEM=VALUE) options in $DATA. Fixed :)
486 - If star notation was used in the scm config file section [inits],
487 [upper_bounds] or [lower_bounds] there was a risk of defining
488 relations that was not specified in [test_relations]. Which in turn
489 would cause a lot of diffrent errors. Fixed :)
491 - Using low_cutoff and hi_cutoff in data::sd resulted in PsN trying to use
492 the new compact data strings as the old arrays. fixed.
494 - Removed unecessary regular expression in output file parser.
496 - Not really a bug. But i tried to add "FIX" as a record option to a theta
497 records. Then when "store_init" was called on the "FIX" option PsN crashed
498 because the option was not a "theta_option". So i added a check for
499 "store_init" abilties... if FIX would become a record option in the future.
501 - Member accessor "datafile" replaced with "datafiles" in model. Affected
502 only the bootstrap which have been updated.
504 - An assumptions about outputfiles crashed sumo occationally. Some more
505 rigorous contentchecking has been added and fixed the problem.
507 Internal changes
508 - Cleanup of dubious utility options.
509 - Option help sorted alphabetically.
510 - "flush", "target" removed from model.
511 - "remove_temp_files" removed from modelfit.
512 - Changed the Makefile considerably, much simpler now.
513 - added "column_to_array" function in data module.
515 Changes from 2.1.9 to 2.1.10
517 Features
518 - The 'picky' option of all utilities now triggers on
519 COVARIANCE STEP ABORTED and
520 PROGRAM TERMINATED BY OBJ
522 Changes from 2.1.8 to 2.1.9
524 Features
525 - Nordugrid support reintroduced (but will probably not be developed further)
526 - Sumo now supports multiple problems and subproblems.
527 - Sumo has new option "precision" which control the number of digits
528 to present estimates with.
529 - An extra 'short' logfile is output from the scm with the most important
530 information.
532 Bugs
533 - Bootstrap option "stratify_on" doesn't work with columns defined in
534 extra data files given with "extra_data" option. Not Fixed.
535 - Resumes of scm's with extra_data_files is now possible. Still some
536 problems with scm's where search_direction=both.
537 - LSF options was not communicated properly between tools and subtools
538 so submission of bootstraps, llp, cdd and scm did not work. Fixed.
539 - Default values for LSF can now be configured in psn.conf.
540 - Print order of modelfiles places $MIX before $PRED. Fixed.
541 - extra_files option was broken. Fixed.
543 Internal Changes
544 - Tool will always fork subtools, this will save memory in the scm (and
545 other recursive tools).
546 - Common options need only be specified once. In one of three categories:
547 model_options, tool_options, script_options.
548 - models can now be created with "reference" objects. This solved the
549 scm resume problem.
550 - Added function to model that cleans extra data code.
552 Changes from 2.1.7 to 2.1.8
554 Features
555 - Rudimentary support for execution on Platform LSF grid systems is now
556 included. The following extra attributes are available to all tools
557 for controlling the execution:
559 run_on_lsf - Use LSF
560 lsf_queue - Use this queue on the grid
561 lsf_options - General options for the grid queuing system.
562 lsf_project_name - Optional. May be used by the grid system
563 administrator to register control project
564 cpu usage.
565 lsf_job_name - Optional. May be used by a grid system
566 administrator to control multiple runs on a
567 grid system.
569 - All tools now have a raw_results file holding the diagnotics and
570 parameter estimates for each NONMEM run.
571 - All tools now handles model files containing one $PROBLEM and no
572 SUBPROBLEMS only.
573 - The Bootstrap and CDD routines now create separate files holding
574 information about which individuals that were included/excluded
575 in/from each created data set.
576 - The Bootstrap results file now has warnings section to better
577 display which runs that were used for result calculations and which
578 were not
579 - PsN now tries to read psn.conf from the users home directory. (Will
580 require some extra documentation).
581 - Optimized modelfit thread operation by removing Storable::Store.
582 - The clean options has four levels of operation.
584 0 - No cleaning at all.
585 1 - Clean more, but make sure resumes work (this is default).
586 2 - Clean more, but make sure summarize work.
587 3 - Clean everything, keeping only results and logs.
589 - Some attributes and accessors of the PsN-Core class 'output' have
590 had their names changed to make it more clear what their purposes
591 are:
592 termination is now called minimization_successful
593 termination_message is now called minimization_message
594 covstep_termination is now called covariance_step_successful
596 these attributes are new:
598 covariance_step_warnings (boolean; 0 or 1)
599 estimate_near_boundary (boolean; 0 or 1)
601 The attribute which was previously called termination could have the
602 values 0, 0.5 and 1 for 'minimization terminates', 'minimization
603 successful but warnings or errors were printed from the covariance
604 step' and finally 'minimization successful'. The new attribute
605 'minimization_successful' can only take the values 0 and 1 and the
606 new attribute covariance_step_warnings now takes the value 1 when
607 covariance step warnings were found. Checks are also perfomed to
608 test whether the estimates are close to any boundary.
609 - Html documents is now generated from command line help text.
610 which ensures that they are consistent and up to date.
611 - Personal configuration file now available under Unix like OS:s
613 Bugs
614 - Options specific to modelfiles were ignored in some Utilties.
615 - Gradients now output in windows. Just like with nmfe.
616 - Bootstrap now consumes considerably less memory.
617 - Removed unused/obsolete "resume" function from modelfit.
618 - Sorted method entries in modelfit diagram.
619 - Fixed bugs in parsing of OFV and eigenvalues.
621 Changes from 2.1.6 to 2.1.7
623 Bugs
624 - "clean" and "compress" options now work under Windows.
625 - SCM now handles crashed runs appropriately.
626 - The ofv_forward, ofv_backward, p_forward and p_backward options of
627 scm now works as they should.
629 Features
630 - SCM improvements. The final model is now available in the
631 first-level scm directory of each step.
632 - The Utilties -help option is improved. You can now give a list of
633 options to -help for specific documentation of those options, like
634 so:
636 $ scm -help do_not_drop
638 - Minor changes in how the [nm_version] section in psn.conf i
639 handled. Look in the psn.conf file that is distributed with PsN for
640 more information.
641 - SDE modelfiles are now supported. (see the -sde_records option)
642 - The update_inits method of the model class now accepts an output
643 file name as input.
644 - Some attributes and accessors of the PsN-Core class 'output' have
645 had their names changed to make it more clear what their purposes
646 are:
647 finalgrad is now called final_gradient
648 parampath is now called parameter_path
649 gradientpath is now called gradient_path
650 covterm is now called covstep_termination
651 - The Utilties now saves the command line that started it in a file
652 "command.txt" under the corresponding directory.
653 - Added a "silent" option to the Utilties that quenches all output.
654 - "nice" option added to utilties that allows you to set priority in
655 unix.
656 - Rudimentary support for NorduGrid reinstated.
657 - Installer script rewritten. It now supports Perl installations in
658 nonstandard directories. Discrepancies between *nix and MSWin32 is
659 smaller.
661 Internal changes
662 - The code that is autogenerated with "dia2code" is optimized a
663 bit and saves a hash "__valid_parameters" in each object, which
664 allows the constructors to be used as a generic (shallow) copy.
666 Misc
667 - continuous spelled correctly throughout SCM.
670 Changes from 2.1.5 to 2.1.6
672 Bugs
673 - Stratify_on bug in bootstrap method fixed.
674 - Data files in directories different from model file are now
675 possible.
676 - Path name handling reimplemented. Relative paths are now handled
677 better.
678 - Models that generate table files would make PsN complain about
679 missing files.
680 - Its now possible to specify multiple output names as a
681 comma-separated list to execute.
682 - The scm can now go backwards.
683 - Handling of table files now responds to the
684 "ignore_missing_output_files" flag.
685 - nmtran error messages was lost in Windows 2K. Execution of nmtran
686 is more plattform independent now. It uses a temporary file, but
687 it is never bigger than the nmtran messages.
688 - Fixed a bug that limited choice of compiler under unix.
689 - Typo in scm utility fixed. (included_relations in backwards step).
691 Features
692 - Added support for "$NONPARAMETRIC" record in model files.
693 - A few improved error messages related to parsing of model files.
694 - The SCM config file now features sections for parameter bounds and
695 user-defined code
697 Internal changes
698 - psn.conf settings are now available as global variables. psn.conf
699 is only parsed once.